8-7-2006
Welcome and hope you find these files of use to you
This zip files includes the following:
SFO FT PSO v1.afl
and FT-Utilities6.afl
12/30/04
Bert
Steele- SFO
that uses FastTrack signals and PSO ...
Here is a version of SFO that uses ...
1. Fred's posted SFO afl file and logic.
2. Bruce's FT-Utilities6 (as posted on FT-Talk) to read FastTrack signals into
the SFO afl.
2. Code written by Bruce that will read the signals once (or only a few times).
This is not my work. Do not thank me. Thank Fred and Bruce. However, if you have
problems with the attached then ask me and blame me.
To make life easy, I have attached the afl, FT-Utilites6.afl and the Settings
file that is needed for SFO.
This SFO does not have all of the features that Fred provided. In the attached
afl, the optimization statements for the following two sections was removed.
// Position size relative to position size up
// Position size relative to position size down
SFO FT PSO
v1optimized.afl
12/31/04
Bert Steele-
Here is the same SFO afl file with
the optimization statements for the following included as commented statements.
// Position size relative to position size up
// Position size relative to position size down
NYBT0.afl
1/1/05
Bert Steele -
The NY Basic Thrust signal, via
NYBT0.ini (that is a zero and not a letter O) works well in combination with
other signals. Reference, if you will, Ultra postings by John Cody, Complite1
and Composite1 postings by Bruce and others, OTCMTP and RUTMTP code enhancements
by Don Bell and others, SFO runs by myself and others.
Here is NYBT0.afl. Here are some attributes.
1. All signals match those provided by Trade
2. AFL code sets 9/1/1988 to Sell consistent with Trade.
3. AFL writes a signal ANYBT0.sig as well as fnu ANYBT0.fnu. The "A" is for
AmiBroker, if you will.
4. AFL uses AMA rather than EMA in an attempt to match Trade to AmiBroker
standards.
5. AFL includes the following lines of code to allow backtesting over different
periods.
Buy = BarsSince(Buy) < BarsSince(Sell);
Sell = BarsSince(Sell) < BarsSince(Buy);
6. AFL is set to allow optimization of parameters. Do in conjunction with PSO.
NYBT0a.afl
1/2/05
Bert Steele-
I posted NYBTO.afl a
few days ago. Bruce provided suggested changes. The attached NYBT0.afl is an
update that includes these changes. Changes include.
1. Bruce's suggested enhancement to allow Lev1 to be less than Lev2 or greater
than Lev2.
2. Corrected the "chicken and egg" problem. I know understand.
3. Added Flip and then ExRem to remove extra signals. There are none, but will
probably use routine as a stardard just to make sure.
NYBVT0.afl
1/2/05
Bert Steele-
NYVT0.sig is another
Thrust signal. It comes from NYBVT.ini and is a NY Volume Thrust indicator.
The Buy and Sell signals match those from Trade EXCEPT when the Trade Stretch
command is used. I am not a Trade expert, but lets assume that Stretch = 8. Then
the signal will stay on a BUY or Sell for a minimum of 8 days.
Stretch Function.afl 1/2/05 Ken Close-
Fuseme.afl
1/9/05
Bert Steele-
Several on the Ultra and Trade Boards have complemented the
fuseme signal. From the attached afl ...
// Concept: When Fuseme is on a Sell, then reduce position size as
// being long has greater risk and, when this plus other "long"
// signal are on a Sell, then shorting has less risk.
The Buy and Sell dates match those provided my fubar5.ini, the source that I
used to create this afl, except for the first signal. This appears to be due to
a difference in the Rsi in AB versus Trade. Consider using in SFO
along with your other signals. This signal goes back to mid 1997. If you
need more data, then you could use Proxy2.ini from the Trade Board (or
Proxy3.ini) to create longer history. I am using PRCGX with correlation at 80
for the entire fusmx history. Not too great, but not bad. I create a new fnu
called FUSMZ when I do this and then I just change the foreign statement in the
afl.
RSISif.afl
1/23/05 Bruce Robinson-
In case anyone is following this and is interested, I'll post the AFL tailored
for Fred's PSO (since what I'm working with isn't user consumable). It is
probably the simplest example of how to have an optimizer turn rules on and off,
and decide what matters. I hope that it will also show one of the possibilities
of the easy to use tool Fred has given you.
A couple of notes one the code, which was done quickly this morning.
1. The boilerplace at the front for the trading parameters is done to allow it
to run on anyone's AB configuration without having to mess with too many
settings. The AA settings will have to be set for long/short, though, and the
symbol and dates set appropriately in the AA window.
2. The program itself was really an exercise in using a good editor and
Copy/Paste/Replace. The buy and sell levels are set up with Optimize()
statements. Each level has an associated "switch" to enable or disable it in the
signal logic. Although the technique is used for all RSI crosses in this case,
it can be used for different signals also.
3. By setting the ranges of the Optimize() statements for the levels, only one
level per RSI decile is allowed. This is arbitrary.
4. The optimization is done on RUT-I and true shorting of RUT-I, while Don's
test was done on a -1x index. It is close enough since the holding periods are
not long.
There are other alternatives to the "switches", but since the optimized 0/1
values are set in the final default value of the Optimize() statement at
completion of optimization, and it makes it easy to see which levels are "on" or
"off".
RSISigb.afl
& RSISig.png
1/26/ 05
Fred Tonneti-Regarding
RSISig.AFL ... Taking a shot at letting the parameter values go where they
seemingly "want" to without restriction results in ...